W3C home > Mailing lists > Public > w3c-sgml-wg@w3.org > June 1997

Re: namespace viz validation [was RE: DSIGS]

From: Henry S. Thompson <ht@cogsci.ed.ac.uk>
Date: Thu, 19 Jun 97 11:32:11 BST
Message-Id: <3403.199706191032@grogan.cogsci.ed.ac.uk>
To: w3c-sgml-wg@w3.org
[David and Paul about validation of documents with namespace-qualified
identifiers]

I THINK there are the pieces of a full proposal on this issue on the
table:

1) I've made a proposal for scoping arbitrary regions within a
namespace.  This allow the 'import' of a DTD or DTD fragment
interpreting all identifiers as being in the namespace, e.g.

<!NS [ html: [ 
<!entity % htmldtd PUBLIC "xxx">
%htmldtd;
]]>

2) I've made explicit something implicit in the Layman/Bray
qualification proposal, which is, to abbreviate, that qualification
inherits.

So if I have after the above

<!element embedhtml (html:html)>
. . .
<embedhtml>
 <html:html>
  <head>...</head>
  <body>...</body>
 </html:html>
</embedhtml>

subsequent to the above, it's valid(atable) and the 'head' and 'body'
are as per the html dtd, by qualification inheritance.

I could of course just use qualified names explicitly in my own DTD,
if I wanted to:

<!element protect (localids:p*)>
<!attlist localids:p id ID #REQUIRED>
. . .
<foo id='p1'>
. . .
<protect>
 <localids:p id='p1'>
  ...
 </localids:p>
</protect>

with no id conflict between the two 'p1' IDs.  The protected one could
be ref'd from outside as 'localids:p1'.

I have foolishly assumed that silence gives consent, and that the ERB
were in due course going to ratify this story, but maybe I'm wrong,
and in any case maybe it doesn't satisfy your requirements.  If so,
please advise.

I certainly do NOT endorse any proposal which says that simply by
qualifying a tag, the enclosing element's type's content model
implicitly gets changed to allow that.  If you want to use e.g. Dublin
Core names and associate your use thereof with Dublin Core, e.g.
(modified Layman syntax)

<?XML-namespace SYSTEM="http:/.../dublincore" SHORT='dc'>

...
<order>
 <dc:date>...</dc:date>
</order
...

then the content model for your 'order' element type better explicitly
include 'dc:core'.

ht
Received on Thursday, 19 June 1997 06:32:19 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 10:04:41 EDT